Wandering albatrosses exploit wind shear by dynamic soaring, enabling rapid, efficient, long-range flight. To explore this flight mode, we compared the ability of a nonlinear dynamic soaring model and a linear empirical model to explain observed variation of the airspeeds of GPS-tracked albatrosses in across-wind flight. In fast winds (> 8 m/s), maximum observed airspeeds reach an asymptote at ~ 20 m/s, whereas the dynamic soaring model predicts much faster airspeeds, up to around 50 m/s. We hypothesize that the birds actively limit airspeed by making fine-scale adjustments to turn angles and soaring heights. Predicted dynamic soaring airspeeds do not extend down to the slowest winds (< 3.2 m/s) of observed flight. We hypothesize that in slow winds wandering albatrosses obtain additional energy from updrafts over water waves. The dynamic soaring model predicts that the minimum wind speed necessary to support dynamic soaring at a cruise airspeed of 16 m/s is 3.2 m/s, achieved via a flight trajectory of linked 137° turns. In reality, observed turn angles are typically ~ 60°. Our simulations suggest that birds may necessarily use smaller turns angles than the theoretical optimum for fast flight in order to limit aerodynamic force on their wings.

Forty-six wandering albatrosses breeding on Bird Island, South Georgia were tracked by GPS during foraging trips made between February to September 2004. GPS positions were recorded at intervals of 0.5–2.0 hours. In addition, activity loggers recorded saltwater immersion so that time actually spent flying could be estimated. We analyzed only direct, sustained, bouts of flight and calculated ground velocity between GPS locations, resulting in 883 velocity measurements.

We obtained wind data from the European Center for Medium-Range Weather Forecasts (ECMWF) on a grid of approximately 125 km in latitude by 75 km in longitude. Wind speed was estimated by interpolation to each bird location and reduced to a reference height of 5 m above mean sea level. This is the median flight height for albatrosses observed from Bird Island and was calculated assuming a logarithmic wind profile and a scale height of 0.03.

The ground velocity of a bird is the vector sum of its average velocity through the air (air velocity) and leeway velocity, which is defined as the bird’s advection by the wind in the downwind direction. Airspeed, the magnitude of air velocity, was calculated by subtracting leeway velocity from ground velocity measured by GPS. Leeway velocity was estimated using the slope parameter of a linear model of ground speed versus the component of wind velocity in the direction of flight Wcos(theta), where W is the wind speed at a reference height of 5 m and theta is the angular difference between wind velocity and ground velocity. We calculated leeway velocity as being equal to the slope parameter (0.51) times wind velocity (see Supplemental Information for details).

We include a data file, a metadata file, and a README file grouped together as part of the Wandering albatross flight dataset.